Sample size and power determination for clustered repeated measurements

Area of Science:

• Biostatistics
• Epidemiology
• Clinical Research Methodology

Background:

• Repeated measurements within subjects are common in epidemiological and clinical studies.
• Subjects are often grouped into clusters, introducing additional correlation structures.
• Accurate sample size and power calculations must account for both within-subject and between-cluster correlations.

Purpose of the Study:

• To develop procedures for computing sample size and power for clustered repeated measurements.
• To provide explicit formulae for comparing means, slopes, and proportions in such designs.
• To facilitate robust statistical planning in complex observational studies.

Main Methods:

• Utilized generalized estimating equations (GEE) for statistical modeling.
• Derived explicit formulae for sample size and power calculations.
• Considered several simple correlation structures for intra-subject and intra-cluster dependencies.

Main Results:

• Developed practical procedures for sample size and power determination.
• Provided analytical solutions for comparing two means, two slopes, and two proportions.
• Demonstrated the importance of accounting for clustered and repeated measures in power analysis.

Conclusions:

• The proposed methods enhance the accuracy of sample size and power calculations in clustered, longitudinal studies.
• These procedures support more efficient and reliable study designs in epidemiology and clinical research.
• The derived formulae offer valuable tools for researchers dealing with complex correlation structures.